econstor
A Service of
zbw
Make Your Publications Visible.
Leibniz-Informationszentrum
Wirtschaft
Leibniz Information Centre
for Economics
Lora, Eduardo; Fajardo, Deisy Johanna
Working Paper
Latin American Middle Classes: The Distance
between Perception and Reality
IDB Working Paper Series, No. IDB-WP-275
Provided in Cooperation with:
Inter-American Development Bank, Washington, DC
Suggested Citation: Lora, Eduardo; Fajardo, Deisy Johanna (2011) : Latin American Middle
Classes: The Distance between Perception and Reality, IDB Working Paper Series, No. IDBWP-275
This Version is available at:
http://hdl.handle.net/10419/88953
Standard-Nutzungsbedingungen:
Terms of use:
Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen
Zwecken und zum Privatgebrauch gespeichert und kopiert werden.
Documents in EconStor may be saved and copied for your
personal and scholarly purposes.
Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle
Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich
machen, vertreiben oder anderweitig nutzen.
You are not to copy documents for public or commercial
purposes, to exhibit the documents publicly, to make them
publicly available on the internet, or to distribute or otherwise
use the documents in public.
Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen
(insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten,
gelten abweichend von diesen Nutzungsbedingungen die in der dort
genannten Lizenz gewährten Nutzungsrechte.
www.econstor.eu
If the documents have been made available under an Open
Content Licence (especially Creative Commons Licences), you
may exercise further usage rights as specified in the indicated
licence.
IDB WORKING PAPER SERIES No. IDB-WP-275
Latin American Middle
Classes:
The Distance between Perception
and Reality
Eduardo Lora
Johanna Fajardo
December 2011
Inter-American Development Bank
Department of Research and Chief Economist
Latin American Middle Classes:
The Distance between Perception
and Reality
Eduardo Lora
Johanna Fajardo
Inter-American Development Bank
Inter-American Development Bank
2011
Cataloging-in-Publication data provided by the
Inter-American Development Bank
Felipe Herrera Library
Lora, Eduardo.
Latin American middle classes : the distance between perception and reality / Eduardo Lora, Johanna
Fajardo.
p. cm. (IDB working paper series ; 275)
Includes bibliographical references.
1. Middle class—Latin America. 2.Latin America—Economic conditions. 3. Income—Latin America. I.
Fajardo, Johana. II. Inter-American Development Bank. Research Dept. III. Title. IV. Series.
http://www.iadb.org
Documents published in the IDB working paper series are of the highest academic and editorial quality.
All have been peer reviewed by recognized experts in their field and professionally edited. The
information and opinions presented in these publications are entirely those of the author(s), and no
endorsement by the Inter-American Development Bank, its Board of Executive Directors, or the countries
they represent is expressed or implied.
This paper may be freely reproduced.
Abstract 1
The main contribution of this paper with respect to previous work is the use of
data on subjective perceptions to identify the Latin American middle classes. This
paper provides a set of comparisons between objective and subjective definitions
of middle-class using data from the 2007 W orld Gallup Poll. Seven objective
income-based definitions of social class are contrasted with a self-perceived social
status measure. Mismatches between the objective and the subjective
classification of social class are the largest when the objective definition is based
on median incomes. Mismatches result from the fact that self-perceived social
status is associated not just with income, but also with personal capabilities,
interpersonal relations, financial and material assets, and perceptions of economic
insecurity. Objective definitions of the middle class based on absolute incomes
provide the lowest mismatches and the most accurate differentiation of the middle
class from other classes.
JEL Classification: D3, I3, D6
Keywords: Middle class, Social status, Income distribution, Latin America
1
Chief Economist and General Manager a.i., Research Department, Inter-American Development Bank. Email:
[email protected]. Research Fellow, Research Department, Inter-American Development Bank. Email:
[email protected]. We acknowledge useful comments obtained from anonymous referees and participants of
seminars at the OECD, IDB, and LACEA 2010 where previous versions of this paper were presented.
1
1. Introduction
Definitions of the middle class used in the economic literature are mainly based on obj ective
measures that classify as such the group of people who are neither at the top nor at the bottom of
the distribution of a statistically measurable characteristic such as income or consumption.
However, as these definitions often rely on arbitrary boundaries around measures of central
tendency, quantiles of the distribution or absolute thresholds, there is little agreement on what the
middle class is. In addition, the economic literature has ignored that social class also refers to
social status, meaning place in a social hierarchy on the basis of life opportunities, life-styles and
attitudes. Sociologists (Hodge and Treiman, 1968; Jackman and Jackman, 1982; Wright and
Singelmann, 1982) argue that no consideration of social class is complete without taking into
account the perceptions of individuals, as these may not coincide completely with their objective
class position but are likely to affect their behavior and choices.
In the context of today’s Latin American countries, social class should be understood as
both a subjective and an economic phenomenon that is the result of a more dynamic social
mobility fostered by increases in income per capita and changes in people’s subjective
interpretation of their class position and of their aspirations.
This paper has two objectives. First, it aims to identify which objective definitions are
closest to a subjective classification of middle-class status by exploring different income-based
measures of social class and their association with a self-perceived social ranking. Since the
mismatches between the objective and the subjective classifications are fairly large, the second
objective of this paper is to explore what factors, in addition to income, are associated with the
self-perceived social ranking of Latin American households.
The remainder of this paper is organized as follows. Section 2 presents a brief literature
review of the concept of middle class. The data source is introduced in Section 3. In Section 4,
we present alternative measures of the objective middle classes and their matching with the selfclassification. Section 5 explores the correlates of self-perceived social status and their ability to
identify self-perceived social classes. The main conclusions are summarized in Section 6.
2
2. What Is the Middle Class?
Social class is a concept long studied in the fields of sociology and economics. The sociological
approach goes back to Marx and Weber’s works on s ocial stratification in the emerging
industrial societies of the nineteenth and early twentieth centuries. Karl Marx defines social
classes on the basis of their distinctive relationships to the means of production and property
ownership: the capitalist class and the working class (Gilbert, 2008). From another standpoint,
Max Weber adds occupation, educational qualifications and life chances for upward mobility to
the Marxist theory of social class based on property. The Weberian theory also contributes to the
study of social stratification by making a clear distinction between class and status. A social
class, an objective economic fact, is a group of people shaped by a similar relationship to the
production and acquisition of goods, i.e., people who share the same economic opportunities but
who are not aware of their common situation and lack class consciousness. In contrast with
social class, status is a subjective concept, a ranking by social prestige and styles of life (Gilbert,
2008).
In economics, there is a large literature on defining the middle class. The middle class is
broadly defined as the group of people who are neither at the top nor at the bottom of the
distribution of a particular indicator, such as a statistically measurable characteristic like income
or consumption. These definitions rely on the (ad hoc) definition of boundaries and, in general,
we identify five main groups of objective definitions: i) definitions based on pe rcentiles, ii)
definitions based on measures of central tendency, iii) definitions based on absolute thresholds,
iv) definitions based on mixed measures, and v) endogenous definitions.
The definition of social class based on percentiles of the income distribution usually
classifies as poor those individuals belonging up to the first two deciles or up to the first, or even
the second, quintiles, and as rich those individuals belonging to the top decile or quintile. The
middle class is, therefore, the group of individuals belonging either to deciles third to ninth
(Solimano, 2008), or to the three middle quintiles (Easterly, 2001; Foster and Wolfson, 2009), or
to the third and fourth quintiles (Alesina and Perotti, 1996). However, as pointed out by Cruces,
López-Calva and Battiston (2010), measures based on quintiles of the income distribution do not
permit analyzing the trend of the middle class size, as this definition is insensitive to changes in
the distribution of income over time.
3
Definitions based on measures of central tendency such as the mean or the median
typically identify the lower bound as a fraction of this measure, whereas the upper bound is
defined as a multiple of the same central tendency measure. For example, Birdsall, Graham and
Pettinato (2000) define the middle class as households in a range between 0.75 and 1.25 times
the median of the household per capita income distribution. Similarly, Davis and Huston (1992)
posit a 0.5 to 1.5 range around the median, and Blackburn and Bloom (1985) define a range of
0.6 to 2.25. In contrast with definitions based on quantiles, definitions based on m easures of
central tendency are sensitive to changes over time in the distribution of income within countries.
This advantage allows researchers to analyze the evolution of the size of the middle class, as
noted by Cruces, López-Calva and Battiston (2010).
While the previous definitions are based on within-country relative incomes, Milanovic
and Yitzhaki (2002), Banerjee and Duflo (2008) and Ravallion (2009) have used absolute
income-based measures to define middle classes. This approach uses an absolute threshold (PPP
adjusted) to divide national income distributions on the basis of the worldwide income
distribution. For instance, Milanovic and Yitzhaki (2002) use the average per day incomes of
Brazil and Italy (12 dollars and 50 dollars, 2000 PPP prices, respectively) to classify as middle
class those households with incomes between these two benchmarks. Banerjee and Duflo (2008)
identify as middle class those households with consumption levels between 2 dol lars and 10
dollars per day or in some cases, from $2 to $4 a day or $6 to $10 day, while Ravallion (2009)
defines the middle class as those with income ranges between 2 dollars (the median value of the
poverty line in 70 developing countries) and 13 dollars (the poverty line in the US) a day at 2005
PPP prices.
A fourth strand of applied work (Birdsall, 2010; Sosa Escudero and Petralia, 2010)
defines the middle class based on a mixed threshold. In particular, Birdsall (2010) defines middle
class in the developing world as people at or above the equivalent of $10 a day, PPP adjusted,
and at or below the 95th percentile of the income distribution in their own country. The first
boundary, $10 a day, is a global threshold below which people are deemed too poor to be middle
class in any present-day global society. Birdsall (2010) proposes the lower boundary as a
reasonable minimum level of economic security that allows households to care about and save
for the future. The second boundary, the 95th percentile, is defined as a local threshold above
4
which people are rich in their own society. In the same study, Birdsall observed low-income
countries in which household income per capita at the 90th percentile was below $10 a day.
Endogenous definitions of middle class have also been proposed (D’Ambrosio, Muliere
and Secchi, 2002; Zhu, 2005; Olivieri, 2008; Massari, Pittau and Zelli, 2009; Cruces, LópezCalva and Battiston (2010). For instance, Cruces, López-Calva, and Battiston (2010) have
proposed a non-parametric definition. For a s ample of Latin American countries, the authors
develop a polarization-based measure that results in a less volatile middle class size over time
and that accounts for greater homogeneity within groups and larger differences between groups
in terms of socioeconomic characteristics.
Solimano (2008) and Banerjee and Duflo (2007) also provide some characteristics of the
middle classes in their studies. In particular, Solimano investigates the relationship of some
variables of economic and political nature with the middle class and finds that middle and higher
per capita income countries have, on average, a larger share of the middle class than low-income
countries. The same effect is observed in countries with lower inequality of income and wealth,
larger governments and a large size of the Small and Medium Enterprises (SME) sector.
On the other hand, Banerjee and Duflo (2008) analyze the patterns of expenditures of the
middle class and find that the share of budget spent on food falls with increases in the standard of
living. Contrariwise, expenditures on education, health and domestic infrastructure increase.
Middle-class consumers typically look for better health care and more expensive education for
their children. Banerjee and Duflo also find that occupational patterns of the middle-class, as
well as their entrepreneurial investments, are similar to those of the poor.
3. Data
Our main data source is the 2007 World Gallup Poll, a survey conducted in 134 countries, which
provides the most extensive coverage of both objective and perceived conditions of quality of
life, including economic and social conditions. The study sample is representative of the
population aged 15 or over in each country. In this study, we use information on 16 Latin
American countries for which the poll provides data on income brackets, which allows us to
construct the household income variable and estimate an income-based definition of middle class
status.
5
Information on i ndividual income levels is not accurately reported in the 2007 W orld
Gallup Poll, but it includes a question on monthly total household income before taxes that is
reported in brackets. However, this question is not always answered by the person who best
knows the income of the household, as the respondent is a randomly selected household member
older than 15. It is also important to note that brackets are expressed in local currency units and
therefore differ across countries.
In their assessment of the Gallup data for the Latin America and Caribbean region,
Gasparini et al. (2009) approximate the household income distribution per country, using
information from household surveys to estimate the intra-bracket distribution, by assigning
random income values in the corresponding income bracket expressed in local currency units.
These values are then converted into US dollars using country exchange rates adjusted for
purchasing power parity (PPP). Since the data set has information on the number of household
members, but not their ages, it permits calculation of per capita household income but not a
household income variable adjusted for the demographic composition of the household. In this
paper, we use the same dataset as Gasparini et al. (2009).
The main advantage of the Gallup Poll is that it allows for international comparisons.
Gasparini et al. (2009) compared the income distribution estimated with the Gallup data and the
income distribution obtained from household surveys, and found them very similar. Although
they found that in the Gallup Poll the poorest and richest quintiles are somewhat smaller than in
household surveys (while the proportion of households in the fourth quintile is in general larger),
the income distribution ranking of countries of the two sources is similar.
4. Measures of the Size of the Latin American Middle Class
As mentioned above, the middle class can be defined objectively on t he basis of some per capita
income thresholds. In this paper, we analyze four different groups of income-based definitions of
middle class, as summarized in Table 1.
Given each definition of middle class, some descriptive statistics can be drawn from our
sample (see Figure 1). The median monthly income per capita for the middle class ranges from
US$358 (in PPP terms) in Ecuador to US$532 in Costa Rica when using the mixed-threshold
definition proposed by Birdsall (2010), while it ranges from US$128 in El Salvador to US$211
in Costa Rica using Ravallion’s (2009) definition. In comparison with other measures, the
6
middle class, as defined by belonging to the 2nd to 8th deciles (Easterly, 2001), has an income
per capita that ranges from US$103 in Peru to US$298 in Argentina. Table 2 presents the middle
class size calculations by country. The different objective income-based definitions show that the
middle class represents from 45 percent in Argentina to 73 percent in Honduras when using
measures based on absolute thresholds; from 18 percent in the Dominican Republic to 46 percent
in Argentina when using measures based on t he median income; and from 7 percent in El
Salvador and Peru to 45 percent in Argentina when using the mixed-threshold measure.
In contrast, we propose a subjective definition of middle class status based on the selfvaluation of relative wealth. By asking individuals what they perceive as their relative wealth
position (on a s cale from 0 to 10), 2 the 2007 Gallup World Survey allows us to get an
approximation of the size and characteristics of the subjective middle classes in Latin America.
On average, Latin Americans rate their relative wealth condition at 4.2.
We avoid the simplest option of considering as subjective middle class a central range of
the ladder question on subjective wealth or any other ad-hoc threshold. Instead, we propose a
definition of subjective social classes that is interrelated with the sizes of the objective classes.
For instance, in our measure of subjective social-classes, we group households in a subjective
middle-class having the same size (by percentage of observations) as an objective middle-class
specified by an income-based definition of class. The procedure is as follows. Firstly, we
generate uniformly distributed random values on a r ange +/-0.5 to translate the categorical
question of wealth condition into a continuous variable for all the individuals in our sample.
Secondly, we rank people from the lowest to the highest value of this continuous variable and
classify the lowest as subjective poor until the subjective-poor group size equals the objectivepoor group size in their respective country. We repeat the second step to classify the following
individuals into subjective middle-class and rich, using the objective middle-class size and the
objective rich-group size as references. As a r esult, for each class within a given country, the
corresponding objective and subjective measures have approximately the same relative size. The
procedure is arguable; however, we benefit from not imposing ad hoc criteria.
The computations reveal that the objective and subjective definitions of middle class only
partly match each other and that mismatches are observed along the whole income distribution.
2
The ladder question on subjective wealth in the Gallup Poll is “Please look at this card. Imagine on one end are
located the ‘Richest people’ of [COUNTRY] and in the other end are located the ‘Poorest people’ of [COUNTRY].
Taking into consideration your current personal situation could you please tell me in which cell you place yourself?”
7
Panel a) in Table 3 displays the percentage of people by income decile that assesses their
standing in each of the 10 rungs of the subjective ladder question. On average, most of the
poorest and the richest individuals believe they belong to the lower-middle fraction of the wealth
distribution in their countries. We also use our definition of subjective social class to estimate the
mismatches. In panel b), we decompose the whole sample of individuals by objective and
subjective classes with one of our alternative objective definitions of middle class: people living
on more than US$2 and less than US$13 a day. By construction, the sizes of the classes are the
same in the objective and the subjective classifications 3 (roughly 17 percent, 66 percent and 17
percent for the poor, middle and rich classes, respectively). However, those that are classified
consistently (and therefore, are placed on the NW-SE diagonal of the table) represent only 57.9
percent of the total sample. Among those in the middle class, 45.5 percent are consistently
classified as such in the objective and the subjective scales. Panel c) in Table 3 displays the
cross-social class classifications in which the middle class are all those individuals with incomes
between 0.5 and 1.5 times the median income of their respective countries. In contrast with the
definition proposed by Ravallion, those that are classified consistently represent only 44.3
percent of the total sample, and around 19 percent are consistently classified as middle class in
the objective and the subjective scales. On average, the inconsistency between objective and
subjective social class has its origins, according to sociologists, in the imperfect correlation
among income, occupation, education and some other factors such as local economic conditions,
employment status, gender, marital status, talent, and luck that create class ambivalence (Hout,
2008).
We can use a matching coefficient as criterion to compare the alternative objective
definitions of middle class. Our matching coefficient corresponds to the percentage of correct
subjective and objective classifications of those belonging to the middle class by country. As
presented in Table 4, definitions based on absolute thresholds and percentiles provide, on
average, matching coefficients between objective and subjective middle-classes that are roughly
similar (62 to 69 pe rcent) and substantially higher than those based on t he median and mixed
measures. This is not only observed in the average for the Latin American countries considered
in this paper, it is also observed in each country (see Table 5).
3
Apart from minor differences, which are less than 1 percent for all countries, due to inability to classify some
individuals by one of the two criteria, as some of them have missing values in either the income variable or the
subjective classification variable.
8
Despite the mismatches, there is a positive and significant degree of correlation between
each objective measure of social class and the subjective definition. For the whole sample, the
Kendall 4 correlation coefficients of the relationship between subjective and objective social
classes displayed in Table 5 confirm that ranking by income is indeed relevant in the subjective
valuation that individuals make of their relative wealth condition; however, the fact that those
coefficients are consistently below 0.3 indicates that there are other factors affecting this
valuation.
5. Characteristics of the Latin American Middle Class
5.1 Correlates of Perceived Social Position
Apart from income, what other factors seem to influence how people see themselves along a
relative wealth scale within their countries? Answering this question may provide a useful
characterization of the subjective middle classes in Latin America. In order to identify what
factors people consider when ranking their social status, we posit that perceived social status
depends on all forms of wealth, real and perceived. Following the classification proposed by the
Inter-American Development Bank (IDB, 2008), those factors can be organized into three main
categories: i) capabilities; ii) relational goods, which include family conditions and other
interpersonal conditions; and iii) material conditions of life, which comprise income, financial
circumstances and physical assets.
The first category, capabilities, includes variables that are specific to the individual such
as gender, age, health status (which can be measured by the EQ-5D, a standardized instrument
that inquires about the presence of health problems in five dimensions: mobility, self-care, usual
activities, pain/discomfort, and anxiety/depression 5) and education level. Capabilities are
necessary conditions for personal fulfillment and social development (Sen, 1985).
The second category, relational goods, is the group of variables referring to the
individual in relation to others. It includes family conditions, such as marital status and
childbearing, and other interpersonal conditions, which reflect the extent and depth of relations
4
Kendall’s rank correlation provides a distribution-free test of independence and a measure of the strength of
dependence between two variables and the similarity between two different orderings.
5
The European Quality of Life-5 Dimensions Index (EQ-5D) is an indicator calculated on the basis of answers to
quasi-objective questions of basic individual health conditions. The original EQ-5D studies were conducted in the
United Kingdom and then implemented in the United States. See Dolan (1997) and Shaw, Johnson and Coons
(2005).
9
of the individual, including whether he declares to have friends to rely on, w hether or not
religion is important in his personal life, being employed and having a supervisor.
The material conditions of life are subdivided into three groups: i) income, ii) financial
circumstances and iii) physical assets. Household income per capita is the most obvious
manifestation of wealth. If all forms of wealth were adequately measured through the other
variables considered in our model, and if all of them had perfect functioning markets, it would be
unnecessary to include income separately in the regression, as total income would correspond to
the flow of returns from all forms of wealth. Since these conditions are not met, the inclusion of
income is clearly warranted.
The influence of income on w ealth perception suffers from endogeneity, particularly
because both variables could be jointly determined by a set of common variables, like national
economic conditions. However, the problem is ameliorated by including a set of variables on
“perceptions of financial circumstances,” since these should capture the influence that temporal
shocks may have on both income and the subjective perception of wealth.
Financial circumstances comprise real and perceived circumstances. To summarize the
information on access and use of financial services, we have constructed an Access to Financial
Services Index, calculated with Principal Components Analysis, PCA. For its calculation we
include the following list of dichotomous variables: whether or not the individual has savings
account, checking account, ATM card, certificates of deposit, credit card and savings for
retirement. Perceptions of financial circumstances may affect how people see themselves along
the wealth ladder. They are measured with the answers to the questions of whether or not the
individual experiences shortages of income to cover food and housing costs, and a composite
variable that summarizes the absence of other financial concerns. 6
Finally, physical assets include variables of ownership of non-financial assets such as
house, television, computer, automobile, washing machine, freezer and house. We also include in
this subgroup variables of access to running water and electricity as well as the location (urban or
rural) of residence as proxies of the possession of, or access to, other assets.
6
A household head is considered to have financial worries if he states that he faces one or more of the following
problems: i) not being able to pay for children’s education, ii) fears of not having enough money for retirement, iii)
not being able to maintain his standard of living, or iv) not being able to afford the medical costs of a serious illness
or accident. The composite variable of not having financial concerns was calculated using the Principal Components
Analysis methodology.
10
To estimate the correlates of subjective social status we implement an ordered logistic
regression analysis on t he ladder question of relative wealth condition. 7 To summarize our
findings we follow the results presented in Table 6. In addition, we evaluate the robustness of our
estimation by including some psychological traits variables as regressors in the estimation of
correlates of perceived wealth (see column 2). 8 The main conclusion of our findings is that
people judge their relative wealth condition taking into consideration all forms of capital, not just
their current income.
First, individuals’ judgment of their relative wealth position is affected by their human
capabilities. Women might tend to conform more than men, as they are more likely to place
themselves in the higher rungs of the ladder. Age shows the familiar U-shape found in happiness
studies, which in this context implies that, controlling for income and all the other factors
mentioned, self-classification in a wealth ladder declines with age until about 72 years of age,
and then increases. Although no definite explanation has been given for this pattern, it could be
associated with changes in aspirations. 9 Notice that this pattern could not be the result of life
cycle factors, since this would imply an inverse U-shape, whereas income and wealth tend to
increase with age until about retirement age, and then decline. Other aspects of human
capabilities that influence perceived wealth status are health status and education, which is
entirely consistent with the hypothesis that human capital is part of subjective wealth.
The same goes for the different forms of relational capital, which are sources of interpersonal relations and support, such as family, friends and religion. Thus, having a spouse and
having children are associated with a higher subjective classification. Surprisingly, being
divorced, as compared with being single, is also associated with higher subjective social status.
In this regard it should be noted that that our estimates point only to correlates of subjective
social status, without implying causality (divorce may be more common among those with more
7
We have included country dummies in the regressions to control for differences in asset prices and other important
unobservable country-specific characteristics.
8
These psychological traits are reflected in a set of subjective wellbeing indicators. Gallup asked individuals about
their life attitude using the following question: “There are all sorts of attitudes towards life. Of those listed, which
one comes closest to your own personal attitude? (Single Response) Work hard and get rich / Live each day as it
comes, cheerfully and without worrying / Don’t think about money or fame, live a life that suits your own tastes /
Resist all evils in the world and live a pure and just life / Try to make a name for yourself / Never think of yourself,
give everything in service to society.”
9
Blanchflower and Oswald (2004) suggest that, in order to explain the U-shaped curve in wellbeing, “one
possibility is that individuals learn to adapt to their strengths and weaknesses, so in mid-life quell their infeasible
aspirations.”
11
wealth, but may not necessarily be a source of higher subjective social status). Having friends is
also associated with higher subjective wealth, since they may be a source of help and support.
However, within each social class friendship may mean something different. Psychological
research (Argyle, 1994) shows that the poor tend to choose as friends people who they can
always turn to for help (mainly, their families) whereas the middle-class describe friends as
people whose company they enjoy. Argyle (1994) also points out that people prefer to choose
friends who are from the same social class or occupational group, and that this tendency is
stronger at the top and bottom of society (middle-class people deliberately make friends from
different settings). We also observe a similar effect of religion on the subjective classification of
Latin Americans.
Material conditions of life are, of course, central in how people judge their relative
standing in society. Income is a strong determinant of subjective social ranking, as mentioned
above. Our estimates imply that when income doubles, keeping everything else constant, the
probability of being at the sixth rung of the wealth ladder of the subjective wealth scale increases
by 1.18 p ercentage points. 10 Apart from income, however, many other aspects of the financial
and material situation of individuals affect their self-evaluation of relative wealth. Having access
to financial services and ownership of a variety of physical assets certainly contributes to feeling
richer. Perceived social status is strongly associated with feelings of economic vulnerability (as
captured in the variables “not having shortage of income to cover food,” “not having shortage of
income to cover housing costs,” and “not being concerned with financial matters”). These results
are in line with the findings of Solimano (2008), who found, under certain circumstances, a
positive correlation between the size of the middle class with the country’s income per capita and
the level of net wealth composed of physical and financial assets, housing, and debts. Finally, the
results in the second column of Table 6 show that our results remain unchanged after controlling
for psychological traits that might bias respondents’ views of their social ranking.
10
By way of comparison, using the same Gallup dataset and the question “On what step of the ladder do you feel
currently, with the highest step (10) representing the best possible life for you and the lowest step (0) representing
the worst for you?”, the ceteris paribus effect of doubling income implies that the probability of being at the sixth
ladder in the life-satisfaction 0-10 scale increases by 0.37 percentage points.
12
5.2 A Detour: Valuing the Different Components of Subjective Wealth
The relative importance of each type of capital in people´s assessments of their own wealth can
be assessed using the so-called life satisfaction approach originally developed by Frey,
Luechinger and Stutzer (2004) to value public goods. These authors use subjective well-being
regression results to assess individuals’ preferences for public goods or externalities. For this
purpose, the public good in question is included as an additional explanatory variable in the
micro-econometric “happiness function,” where a measure of happiness or life satisfaction is the
dependent variable. The estimated coefficient can be interpreted as the marginal utility of the
public good. Together with the estimate for the marginal utility of income, the marginal rate of
substitution between income and the public good can be calculated, thus providing the valuation
(in terms of income) of the public good. The life satisfaction approach has been used to estimate
the value of public goods and externalities such as air quality (Luechinger, 2009) and terrorism
(Frey, Luechinger and Stutzer, 2009), as well as several personal capacities and interpersonal
conditions that contribute to life satisfaction (IDB, 2008). The same method of appraisal can be
used to calculate the income-equivalent values of all the forms of capital that contribute to
people’s subjective valuation of their own wealth. Column 3 of Table 6 presents the valuations of
all the significant correlates in regression 1 of the same table.
This valuation requires the estimation of the marginal effects at particular values for the
independent variables. In order to specify such values, we have arbitrarily chosen a hypothetical
individual, more specifically a 30-year-old Brazilian woman, in (self-reported) perfect health,
who has completed higher education, who is married with one child, who considers religion to be
important in her life, has friends she can trust, and is employed and has a work supervisor. All
these features are desirable in the sense that they are associated with higher levels of subjective
well-being, according with regression 1. Per capita income in this person’s household is assumed
to be US$157 monthly (in PPP terms), which corresponds to the median income for the whole
sample of Latin American countries. We further assume this individual to have all the desirable
financial and material conditions that are associated with a h igher self-reported wealth level
(namely, access to financial services, no financial concerns, house ownership, possession of all
basic housing services and assets, and residence in an urban area).
Then, we calculate the income change required to keep constant the subjective wealth of
this individual (which we have assumed to be 6 on the 0 to 10 scale) if this person did not have
13
the desirable characteristics enumerated above. For instance, we find that if the health status of
this woman were to deteriorate to that of the 25th percentile with worst health within Brazil, the
loss in terms of subjective wealth would be equivalent to 0.57 t imes her income. In a similar
way, if she did not have complete higher education, but only complete secondary education, the
income equivalent loss of subjective wealth would be 2.3 t imes her income. By adding the
corresponding valuations of the three possible education levels, it can be concluded that the
income equivalent value of her human capital in education is 4.11 times her income. Therefore
her total human capital (in health and education) is valued at 4.68 times her income.
The value of her relational capital can be computed in a similar way. If she were single
instead of married, had no children, religion were not important in her life and had no friends she
could trust, she would need to have 3.37 times more income in order to have the same subjective
wealth. (Had we assumed that the hypothetical individual originally was divorced and had
several children, this calculation would go up to 3.78 times her income).
Similarly, the income equivalent value of all the desirable financial circumstances
amounts to 5.2 times the income of the individual, and the value of all her material possessions
amounts to 8.26 times her income.
Putting together all the forms of subjective capital of this hypothetical individual, we can
conclude that her capital adds up to 21.51 times her income. This is a stark reinforcement to the
conclusion that income is but one minor component of people’s self-evaluation of their relative
wealth. These calculations could be used to compute the distribution of wealth among
individuals or social classes (a topic that will be pursued in a separate paper).
5.3 Are Subjective Middle Classes Different than the Rich and the Poor?
The importance of the correlates of the subjective social ranking discussed above can be further
tested by assessing whether they help to discriminate effectively between the subjective middle
classes and the other two classes. Our dependent variables will be categorical (belonging or not
belonging to a subjective class), as defined in a previous section. Since we have seven alternative
definitions of the objective middle classes, we also have the corresponding seven subjective
definitions (constructed so that class sizes match as explained above). The issue considered here
is whether the subjective middle classes can be differentiated from the other classes on the basis
14
of the variables associated with the self-ranking of wealth? Results from multinomial logit
regression models are presented in Table 7.
A cursory reading of the results immediately reveals that, across all definitions of the
middle class, there are many more factors that help discriminate between the middle class and
the poor than between the middle class and the rich. While no fewer than 20 variables help to
differentiate the middle class from the poor in four or more of the definitions used, there are just
six variables that contribute to discriminate between the middle class and the rich in four or more
definitions. They are (in the order they appear in the table): have friends, income, shortage of
income to cover food costs, financial concerns, access to telephone service, own computer, and
own automobile. All are discriminators between the middle class and the rich that are also
discriminators between the middle class and the poor for most of the definitions. In contrast, a
long list of factors consistently differentiates the middle class from the poor (that is, across at
least four definitions) while not contributing to differentiating it from the rich. Personal
characteristics include lower age, better health and more education (of all levels). Having just
one child is not a common feature among the poor, and so helps differentiate it from the middle
classes. Assets and possessions that differentiate consistently the middle class from the poor
include owning a house, television and freezer. Living in an urban area is also a feature of the
middle class that differentiates it from the poor.
To simplify these results greatly, a Latin American who defines herself as middle-class is
someone younger, with better health and more education than the poor, who has already
managed to acquire some of the possessions of the rich (including a house, and in a few cases a
computer, but probably not an automobile), and who has more income, more friends and more
financial security than the poor but not as much of all those things as the rich. However, on the
basis of all these variables, it is much easier to tell how the middle class differentiates from the
poor than the rich.
Some of the definitions of subjective middle class lend themselves better to
characterizing who is and who is not middle class in Latin America. The definitions that match
those based on a bsolute income thresholds perform better in the sense that a l arger pseudo Rsquared is found compared with those that match those based on relative incomes. 11 However, a
11
The model predicts correctly middle- class status in around 63 percent of the cases, when the definition based on
the absolute threshold proposed by Ravallion (2009) is used. In comparison with the classification based on
15
pseudo R-squared of just around 0.15 for the absolute threshold-based measures suggests that the
factors identified are vastly insufficient to fully understand the reasons that lead people to see
themselves as middle class.
6. Concluding Remarks
While sociologists have noted that field that proper analysis of social classes must consider both
objective and subjective factors, the economic literature often ignores subjective aspects of class
and opts for social class analysis based on objective variables such as income and consumption.
In this paper we consider both strands of theory and use a subjective definition of social
stratification to compare its match with seven alternative, income-based, statistical definitions of
middle-class in Latin America using the rich dataset of the 2007 World Gallup Poll.
The size of the middle class varies across the different objective definitions of middleclass not only on t he average but also by country. For example, Argentina’s middle class
encompasses 57 percent of the population when it is defined based on an absolute threshold of 2
to 13 dollars a day, but only 46 percent when it is defined based on an interval of 0.5 to 1.5 times
the median country income.
Social classes derived from income-based definitions using absolute thresholds and
percentiles provide, on average, the highest matching coefficients with a s ubjective social
classification based on a ladder question of individual relative wealth. Sociologists argue that
such inconsistency between objective and subjective social class is due to class ambivalence
created by the imperfect correlation among observables such as income, occupation, education
and some other factors such as local economic conditions, employment status, gender, marital
status, talent, and luck.
Several the factors underlying the discrepancies between objective and subjective social
classes are identified in this paper. People consider many variables other than income when
ranking their social status. More precisely, they consider all forms of capital, including personal
capabilities, relational goods and material conditions of life, in their self-assessment of their
position in society. The relative importance that they attach to each form of capital in relation to
income can be used to compute their income-equivalent value.
absolute-threshold measures, this percentage declines to 28 percent when the relative definition proposed by Davis
and Huston (1992) is considered.
16
The same set of factors that is associated with the self-ranking of individuals along the
wealth ladder is used in this paper to identify the distinctive characteristics of the subjective
middle-classes vis-à-vis the self-defined poor and the self-defined rich. For that purpose, we use
alternative definitions of subjective middle class so as to match the size and relative position of
the middle classes according to different objective definitions. Consistently across definitions, it
is much easier to discriminate the subjective middle class from the subjective poor than from the
subjective rich. A Latin American who defines himself as middle class is someone younger, with
better health and more education than the poor, with some of the possessions of the rich, and
who has more income, more friends and more financial security than the poor but not as much as
the rich.
The set of factors considered is more successful in discriminating the middle class from
the other classes when the definitions of subjective middle-class correspond to the objective
definitions based on a bsolute income thresholds (or on a mix of an absolute threshold and a
relative income). Our results show that older people have a greater tendency to self-classify as
poor whereas younger people tend to classify themselves as either middle-class or rich. Women
are less likely to see themselves as poor as compared to men, whereas having some level of
secondary or higher education increases the odds of being self-classified as middle-class or rich.
On the contrary, having financial concerns and feelings of economic vulnerability are more
characteristic of those in the poor and middle classes. Since the objective definitions based on
absolute thresholds also produce some of the best matches with the corresponding subjective
definitions, they should be taken as superior to the alternative definitions based on r elative
incomes or median incomes. In any case, it is clear from our results that income is just one of the
many factors that influence social class self-perception.
17
References
Alesina, A., and R. Perotti. 1996. “Income Distribution, Political Instability and Investment.”
European Economic Review 40(6): 1203-1228
Allardt, E., and H. Uusitalo. 1972. “Dimensions of Welfare in a Comparative Study of the
Scandinavian Societies.” Scandinavian Political Studies. Bind 7.
Available at:
http://www.tidsskrift.dk/visning.jsp?markup=&print=no&id=95959
Argyle, M. 1994. The Psychology of Social Class. First edition. London, United Kingdom:
Routledge Press.
Banerjee, A. and E. Duflo. 2008. “What is Middle Class about the Middle Class around the
World?” Journal of Economic Perspectives 22(2): 3-28
Birdsall, N., C. Graham and S. Pettinato. 2000. “Stuck in the Tunnel: Is Globalization Muddling
the Middle Class?” Center on S ocial and Economic Dynamics Working Paper 14.
Washington, DC, United States: The Brookings Institution.
Birdsall, N. 2010. “The (Indispensable) Middle Class in Developing Countries; or, the Rich and
the Rest, not the Poor and the Rest.” Working Paper 207. Washington, DC, United States:
Center for Global Development.
Blackburn, M., and D. Bloom. 1985. “What Is Happening to the Middle Class?” American
Demographics 7(1): 18-25
Blanchflower, D. and A.J. Oswald. 2004. “Well-being over Time in Britain and the USA”.
Journal of Public Economics 88: 1359-1387.
Cruces, G., L.F. López-Calva and D. Battiston. 2010. “Down and Out or Up and In? In Search of
Latin America’s Elusive Middle Class.” Research for Public Policy, Inclusive
Development, ID-03-2010, RBLAC-UNDP. New York, United States: United Nations
Development Programme.
Cummins, R. 1997. Comprehensive Quality of Life Scale-Adult. Fifth edition. Melbourne,
Australia: Deakin University, School of Psychology.
D’Ambrosio, C., P. Muliere and P. Secchi. 2002. “Income Thresholds and Income Classes.”
Milan, Italy: Universitá Bocconi. Working paper.
Davis, J., and J.H. Huston. 1992. “The Shrinking Middle-Income Class: A Multivariate
Analysis.” Eastern Economic Journal 18(3): 277-285.
Dolan, P. 1997. “Modeling Valuations for Health States.” Medical Care 35(11): 1095-1108.
18
Easterly, W. 2001. “Middle Class Consensus and Economic Development.” Journal of Economic
Growth 6(4): 317-336.
Foster, J., and M.C. Wolfson. 2009. “Polarization and the Decline of the Middle Class: Canada
and the US.” OPHI Working Paper 31. Oxford, United Kingdom: University of Oxford.
Frey, B., S. Luechinger, and A. Stutzer. 2004. “Valuing Public Goods: The Life Satisfaction
Approach.” CESifo Working Paper 1158. Munich, Germany: University of Munich,
Center for Economic Studies.
----. 2009. “The Life Satisfaction Approach to Valuing Public Goods: The Case of Terrorism.”
Public Choice 138(3-4): 317-345
Gallup. 2007. Gallup World Poll. Available at: www.gallup.com/consulting/worldpoll/24046/about.aspx
Gasparini L. et al. 2009. “ Objective and Subjective Deprivation.” In: C. Graham and E. Lora,
editors. Paradox and Perception: Measuring Quality of Life in Latin America.
Washington, DC, Unites States: The Brookings Institution.
Gilbert, D. 2008. The American Class Structure in an Age of Growing Inequality. Seventh
edition. Los Angeles, United States: Pine Forge Press.
Hout, M. 2008. “How Class Works: Objective and Subjective Aspects of Class since the 1970s.”
In: A. Lareau and D. Conley, editors. Social Class: How Does It Work? New York,
United States: Russell Sage Foundation.
Hodge, R., and D. Treiman. 1968. “Class Identification in the United States.” American Journal
of Sociology 73(5): 535-47
Inter-American Development Bank (IDB). 2008. Beyond Facts: Understanding Quality of Life.
Development in the Americas Report. Washington, DC, United States: Inter-American
Development Bank
Jackman, M., and R. Jackman. 1982. Social Class Awareness in the United States. Berkeley, CA:
University of California Press.
Lora, E., and J. Chaparro. 2009. “The Conflictive Relationship between Satisfaction and
Income.” In: C. Graham and E. Lora, editors. Paradox and Perception: Measuring
Quality of Life in Latin America. Washington, DC, United States: The Brookings
Institution
Luechinger, S. 2009. “Valuing Air Quality Using the Life Satisfaction Approach.” Economic
Journal 119(536): 482–515.
19
Massari, R., M. Pittau and R. Zelli. 2009. “A Dwindling Middle Class? Italian Evidence in the
2000s.” Journal of Economic Inequality 7(4): 333-350.
Milanovic, B., and S. Yitzhaki. 2002. “Does Decomposing World Income Distribution: Does the
World Have a Middle Class?” Review of Income and Wealth 48(2): 155-178.
Olivieri, S. 2008. “Debilitamiento de la Clase Media: Gran Buenos Aires 1986-2004.” La Plata,
Argentina: Universidad Nacional de La Plata, Facultad de Ciencias Económicas.
Unpublished MSc. Thesis.
Ravallion, M. 2009. “The Developing World’s Bulging (But Vulnerable) Middle Class.” Policy
Research Working Paper 4816.Washington, DC, United States: World Bank.
Sen, A. 1985. Commodities and Capabilities. Oxford, England: Oxford University Press.
Shaw, J.W., J.A. Johnson and S.J. Coons. 2005. “U.S. Valuation of the EQ-5D Health States:
Development and Testing of the D1 Valuation Model.” Medical Care 43(3): 203-220.
Solimano, A. 2008. “The Middle Class and the Development Process.” Serie Macroeconomía del
Desarrollo 65. Santiago, Chile: United Nations-ECLAC.
Sosa Escudero, W., and S. Petralia. 2010. “I Can Hear the Grass Grow: The Anatomy of
Distributive Changes in Argentina.” CEDLAS Working Paper 0106. La Plata, Argentina:
Universidad Nacional de La Plata, Centro de Estudios Distributivos, Laborales y
Sociales.
Wright, E., and J. Singelmann. 1982. “Proletarianization in the American Class Structure.” In:
M. Burawoy and T. Skocpol, editors. Marxist Inquiries: Studies of Labor, Class, and
States. Supplement to American Journal of Sociology 88.
Zhu, F. 2005. “A Nonparametric Analysis of the Shape Dynamics of the US Personal Income
Distribution: 1962-2000.” Monetary and Economic Department Working Paper 184.
Basel, Switzerland: Bank for International Settlements.
20
Table 1. Definitions of the Middle Class
Definition
Based on the median (p50) of the
income (y) distribution
Based on percentiles of the income
(y) distribution
Based on absolute thresholds
Based on mixed thresholds
a.
b.
Authors
Davis and Huston, 1992
𝐇𝐨𝐮𝐬𝐞𝐡𝐨𝐥𝐝 𝒙 ∈ 𝑴𝒊𝒅𝒅𝒍𝒆 𝑪𝒍𝒂𝒔𝒔 𝒊𝒇 𝒂
0.5 ∗ 𝐷 −1 (𝑝50 ) ≤ 𝑦(𝑥) ≤ 1.5 ∗ 𝐷 −1 (𝑝50 )
Birdsall et al., 2000
0.75 ∗ 𝐷 −1 (𝑝50 ) ≤ 𝑦(𝑥) ≤ 1.25 ∗ 𝐷 −1 (𝑝50 )
Solimano, 2008
𝐷 −1 (𝑝30 ) ≤ 𝑦(𝑥) ≤ 𝐷 −1 (𝑝90 )
Easterly, 2001
𝐷 −1 (𝑝20 ) ≤ 𝑦(𝑥) ≤ 𝐷 −1 (𝑝80 )
Banerjee and Duflo, 2007 b
2 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 10 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦
Birdsall,2010
10 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 𝐷−1 (𝑝95 )
Ravallion, 2009
2 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 13 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦
Definition in terms of the cumulative distribution D(y), nth percentile Pn, and x’s household income y(x)
The authors also specified two alternative segments on their study: $2 to $4 a day and $6 to $10 day.
21
Table 2. The Size of the Middle-Classes by Country (Objective Definitions)
Objective Middle-Class Size
Based on
Absolute
Thresholds
Country
Number of
Observations
Observations
with Valid
Income
Based on
Percentiles
2 to 10 usd 2 to 13 usd
p20 - p80
ppp a day ppp a day
p30 - p90
Based on the
Median
Mixed
Measures
0.5 - 1.5 0.75 - 1.25 10 usd ppp a
times p50 times p50
day - p95
Argentina
1,000
74%
45%
57%
60%
60%
46%
23%
45%
Bolivia
1,000
69%
58%
64%
60%
60%
42%
22%
10%
Brazil
1,038
86%
58%
67%
60%
60%
45%
24%
27%
Chile
1,023
87%
61%
71%
60%
60%
45%
23%
27%
Costa Rica
1,002
68%
49%
60%
60%
60%
42%
28%
41%
Dominican Rep.
1,000
78%
54%
61%
60%
60%
34%
18%
23%
Ecuador
1,061
92%
63%
69%
60%
60%
44%
24%
10%
El Salvador
1,001
74%
62%
66%
60%
60%
43%
23%
7%
Guatemala
1,000
52%
65%
71%
60%
60%
42%
21%
9%
Honduras
1,000
61%
65%
73%
60%
60%
45%
23%
15%
999
85%
59%
66%
60%
60%
40%
20%
17%
Nicaragua
1,000
92%
58%
65%
60%
60%
41%
23%
17%
Panama
1,000
81%
62%
70%
60%
60%
41%
23%
18%
Paraguay
1,000
85%
58%
65%
60%
60%
39%
20%
17%
Peru
1,000
85%
59%
64%
60%
60%
39%
20%
7%
Uruguay
1,004
64%
47%
59%
60%
60%
39%
20%
40%
Latin America *
17,128
73%
58%
66%
60%
60%
42%
22%
21%
Mexico
* Total Number of Observations, Total Percentage of Observations with Valid Income, and Average Middle Class
Size .
Source: Authors’ calculations based on Gallup (2007).
22
Table 3. Classifying Latin American Social Classes
a) Income Distribution and Self-Assessment of Wealth (% of Individuals)
Subjective Wealth
The Poorest
1
2
3
4
5
6
7
8
9
The Richest
1
7.8
12.2
17.6
18.5
17.5
17.2
+
+
+
*
*
2
5.5
8.8
13.9
21.0
19.3
20.8
5.8
+
+
*
*
3
+
7.8
11.2
19.0
20.9
24.1
7.1
+
+
*
*
Decile of the Income Distribution
4
5
6
7
+
+
+
+
5.4
5.7
5.0
+
11.2
9.9
8.7
6.3
18.1
17.5
19.0
13.2
21.9
21.2
21.3
22.9
24.7
25.1
26.2
29.7
8.6
9.9
9.6
12.9
+
5.1
+
6.3
+
+
+
+
*
*
*
*
*
*
*
*
8
+
+
5.3
14.1
21.3
31.2
12.4
7.5
+
*
*
9
*
+
+
13.8
18.9
30.2
15.7
9.4
+
*
*
* Less than 1%
+ Between 1% and 5%
Note: The data in each column add up to 100 percent.
b) Social Class Definition Based on Absolute Threshold of 2 to 13 USD PPP a Day.
Poor
Objective
Middle-class
Rich
Total Subjective
Poor
5.88
9.93
0.98
16.79
Middle-class
10.13
45.53
10.02
65.68
Rich
0.85
10.24
6.44
17.53
Total Objective
16.85
65.7
17.45
100
Subjective
%
c) Social-Class Definition Based On 0.5 to 1.5 Times the Median of the Income Distribution
Poor
Objective
Middle-class
Rich
Total Subjective
Poor
12.98
11.02
5.09
29.09
Middle-class
11.91
18.79
11.05
41.75
Rich
4.61
11.97
12.57
29.16
Total Objective
29.51
41.78
28.71
100
Subjective
%
Source: Authors’ calculations based on Gallup (2007).
23
10
*
+
+
8.4
14.0
30.7
18.8
13.3
6.4
+
*
Table 4. Matching Coefficients between Objective and Subjective Middle-Class
for Alternative Objective Definitions
Definition
Based on the median (p50) of the
income (y) distribution
Based on percentiles of the income
(y) distribution
Based on absolute thresholds
Based on mixed thresholds
𝐇𝐨𝐮𝐬𝐞𝐡𝐨𝐥𝐝 𝒙 ∈ 𝑴𝒊𝒅𝒅𝒍𝒆 𝑪𝒍𝒂𝒔𝒔 𝒊𝒇 𝒂
Average Matching
Coefficient
0.5 ∗ 𝐷 −1 (𝑝50 ) ≤ 𝑦(𝑥) ≤ 1.5 ∗ 𝐷 −1 (𝑝50 )
45%
𝐷 −1 (𝑝20 ) ≤ 𝑦(𝑥) ≤ 𝐷 −1 (𝑝80 )
63%
2 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 10 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦
62%
10 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 𝐷−1 (𝑝95 )
36%
0.75 ∗ 𝐷 −1 (𝑝50 ) ≤ 𝑦(𝑥) ≤ 1.25 ∗ 𝐷 −1 (𝑝50 )
24%
𝐷 −1 (𝑝30 ) ≤ 𝑦(𝑥) ≤ 𝐷 −1 (𝑝90 )
64%
2 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦 ≤ 𝑦(𝑥) ≤ 13 𝑢𝑠𝑑 𝑝𝑝𝑝 𝑝𝑒𝑟 𝑑𝑎𝑦
69%
Source: Authors’ calculations based on Gallup (2007).
24
Table 5. Correlation and Matching Coefficients by Country (Objective Definition)
Correlation a
Based on absolute
thresholds
Country
2 to 10 usd ppp 2 to 13 usd ppp
a day
a day
Based on percentiles
Matching Coefficient b
Based on the median
Mixed
measures
Based on absolute
thresholds
Based on percentiles
Based on the median
Mixed
measures
p20 - p80
p30 - p90
0.5 - 1.5
times p50
0.75 - 1.25
times p50
10 usd ppp a
day - p95
2 to 10 usd
ppp a day
2 to 13 usd
ppp a day
p20 - p80
p30 - p90
0.5 - 1.5 times
p50
0.75 - 1.25
times p50
10 usd ppp a
day - p95
Argentina
0.188 **
0.196 **
0.154 **
0.190 **
0.196 **
0.200 **
0.184 **
51%
62%
59%
62%
47%
23%
52%
Bolivia
0.207 **
0.186 **
0.246 **
0.204 **
0.186 **
0.166 **
0.137 **
60%
67%
63%
64%
45%
25%
16%
Brazil
0.175 **
0.166 **
0.177 **
0.163 **
0.166 **
0.204 **
0.112 **
60%
68%
62%
66%
49%
22%
32%
Chile
0.333 **
0.341 **
0.351 **
0.364 **
0.341 **
0.300 **
0.312 **
68%
77%
67%
69%
52%
28%
40%
Costa Rica
0.253 **
0.280 **
0.234 **
0.164 **
0.280 **
0.264 **
0.227 **
54%
65%
60%
61%
41%
22%
53%
Dominican Rep.
0.315 **
0.292 **
0.312 **
0.280 **
0.292 **
0.303 **
0.307 **
60%
66%
65%
67%
33%
19%
38%
Ecuador
0.298 **
0.267 **
0.307 **
0.310 **
0.267 **
0.398 **
0.230 **
66%
72%
64%
65%
46%
22%
21%
El Salvador
0.250 **
0.218 **
0.264 **
0.210 **
0.218 **
0.244 **
0.189 **
67%
71%
67%
64%
47%
27%
20%
Guatemala
0.177 **
0.174 **
0.171 **
0.168 **
0.174 **
0.159 **
0.149 **
67%
73%
62%
62%
48%
20%
9%
Honduras
0.120 **
0.087 **
0.126 **
0.105 **
0.087 **
0.075 **
0.105 **
66%
72%
60%
61%
45%
27%
24%
Mexico
0.319 **
0.308 **
0.318 **
0.321 **
0.308 **
0.269 **
0.266 **
64%
71%
66%
67%
45%
24%
25%
Nicaragua
0.263 **
0.268 **
0.271 **
0.290 **
0.268 **
0.298 **
0.187 **
63%
70%
66%
67%
46%
28%
26%
Panama
0.176 **
0.147 **
0.171 **
0.158 **
0.147 **
0.143 **
0.121 **
62%
70%
59%
61%
45%
23%
26%
Paraguay
0.274 **
0.257 **
0.278 **
0.274 **
0.257 **
0.247 **
0.278 *
62%
70%
66%
64%
44%
19%
32%
Peru
0.243 **
0.247 **
0.202 **
0.233 **
0.247 **
0.213 **
0.099 **
62%
67%
63%
62%
44%
24%
9%
Uruguay
0.153 **
0.160 **
0.250 **
0.219 **
0.160 **
0.173 **
0.141 **
52%
60%
63%
62%
36%
20%
45%
Average
0.298 ***
0.285 *** 0.244 *** 0.235 *** 0.245 *** 0.237 *** 0.246 ***
62%
69%
63%
64%
45%
24%
36%
a. Kendall’s Tau Coefficient Correlation for all social classes; significant at: ***99%, **95%, *90%
b. Matching coefficient : Full middle class size to average middle class size ratio, by country
Source: Authors’ calculations based on Gallup (2007).
25
Table 6. Factors Associated with Subjective Social Ranking
Ordered Logit Estimation
Subjective social ranking
(wp5722, 0-10 scale)
(1)
(2)
Valuations in
Times of Income
Capabilities
Female
Age (years)
Age squared
Health score (EQ - 5D)
Complete primary education
Complete secondary education
Complete superior education
0.116***
(0.062)
-0.028***
(0.005)
0.000***
(0.000)
0.882***
(0.212)
0.197***
(0.078)
0.374***
(0.111)
0.582***
(0.106)
0.123***
(0.063)
-0.026***
(0.004)
0.000***
(0.000)
0.804***
(0.214)
0.203***
(0.078)
0.384***
(0.112)
0.600***
(0.108)
0.127***
(0.038)
0.185***
(0.079)
0.219
(0.135)
0.138***
(0.047)
0.168***
(0.086)
0.136***
(0.076)
0.348***
(0.072)
0.052
(0.053)
0.005
(0.059)
0.122***
(0.036)
0.188***
(0.086)
0.230***
(0.127)
0.139***
(0.050)
0.163***
(0.090)
0.129***
(0.077)
0.356***
(0.069)
0.035
(0.050)
0.014
(0.062)
0.267***
(0.044)
0.274***
(0.044)
(a)
0.1
0.57
0.79 (b)
1.02 (c)
2.3 (d)
Relational Goods
Married
Divorced
Widowed
Has one child
Has two or more children
Consider religion to be important
Has friends
Has employment
Has a supervisor
0.53 (e)
0.78 (f)
0.58
0.74 (g)
0.57
1.69
Material conditions of life
Income
Household's monthly per capita income, US$ PPP, logs
(Table continued)
26
Table 6. Factors Associated with Subjective Social Ranking (continued)
Subjective social ranking
(wp5722, 0-10 scale)
(1)
(2)
Valuations in
Times of Income
Material conditions of life
Financial circumstances
Access to financial services index
Does not have shortage of income to cover food costs
Does not have shortage of income to cover housing costs
Not concerned with financial matters
0.163***
(0.051)
0.386***
(0.055)
0.235***
(0.091)
0.261***
(0.029)
0.150***
(0.054)
0.394***
(0.056)
0.226***
(0.091)
0.261***
(0.030)
0.26
0.127***
(0.065)
0.252***
(0.092)
0.218***
(0.042)
0.246***
(0.080)
0.233***
(0.048)
0.177***
(0.067)
0.147***
(0.064)
0.232***
(0.069)
0.218***
(0.124)
0.133***
(0.068)
0.243***
(0.097)
0.226***
(0.044)
0.238***
(0.081)
0.222***
(0.050)
0.182***
(0.064)
0.140***
(0.066)
0.237***
(0.072)
0.205***
(0.123)
0.53
1.91
1.07
1.96
Physical Assets
Owns a house
Access to running water service
Access to telephone service
Has a television
Has a computer
Has an automobile
Has washing machine
Has a freezer
Lives in urban area
1.15
0.98
1.12
1.05
0.77
0.63
1.05
0.98
(Table continued)
27
Table 6. Factors Associated with Subjective Social Ranking (continued)
Subjective social ranking
(wp5722, 0-10 scale)
(1)
(2)
Valuations in
Times of Income
Personality Traits
Life attitude: "Work hard and get rich"
0.129
(0.121)
Life attidude: "Live each day as it comes, cheerfully and without
worrying"
0.079
(0.113)
Life attitude: "Don't think about money or fame, live a life that
suits your own tastes"
0.071
(0.127)
-0.151
(0.172)
Life attitude: "Try to make a name for yourself"
Life attitude: "Never think of yourself, give everything in service
to society"
Cut 1
0.029
(0.108)
0.173
(0.356)
1.364***
(0.356)
2.347***
(0.354)
3.438***
(0.354)
4.514***
(0.386)
6.098***
(0.411)
7.091***
(0.450)
8.156***
(0.497)
9.636***
(0.508)
10.339***
(0.583)
Cut 2
Cut 3
Cut 4
Cut 5
Cut 6
Cut 7
Cut 8
Cut 9
Cut 10
Observations
2
Pseudo R
Log Likelihood ln L(β)
0.186
(0.365)
1.383***
(0.360)
2.370***
(0.357)
3.461***
(0.353)
4.538***
(0.387)
6.118***
(0.421)
7.111***
(0.466)
8.178***
(0.520)
9.642***
(0.528)
10.344***
(0.603)
8,613
8,373
0.094
-15564.131
0.094
-15134.656
Significant at: ***99%, **95%, *90%. Country dummies are not reported.
Source: Authors’ calculations based on Gallup (2007).
Notes: a) Individual depicted in this valuation is a married 30-year-old Brazilian woman with one child and high
education level, employed, with friends and religious beliefs, who lives in a house with all public utilities. The
valuation only takes into account significant variables in Column 1. This statement applies wherever not indicated
by a different note. b) This is the value of having only primary education. c) This is the additional value of having
secondary education. d) This is the additional value of having superior education. e) Marriage is compared with
single. f) Divorce is compared with single. g) Having two or more children is compared with not having children.
28
Table 7. Factors Associated with Feeling Middle Class, by Definition
Multinomial Logit Estimation
Dependent Variable:
Subjective social ranking
Based on absolute thresholds
US$2-US$10 PPP a day
US$2-US$13 PPP a day
(1)
(2)
(1)
(2)
Based on percentiles
p20 - p80
p30 - p90
(1)
(2)
(1)
(2)
Based on the median
0.5*p50 -1.5*p50
0.75*p50 -1.25*p50
(1)
(2)
(1)
(2)
Mixed measures
US$10 PPP a day - p95
(1)
(2)
Capabilities
Female
Age (years)
Age squared
Health score (EQ - 5D)
Complete primary education
Complete secondary education
Complete superior education
-0.073
(0.075)
0.032***
(0.016)
-0.000***
(0.000)
-0.814***
(0.201)
-0.258***
(0.089)
-0.499***
(0.103)
-0.702***
(0.147)
0.077
(0.101)
-0.014
(0.011)
0.000
(0.000)
0.625***
(0.280)
-0.066
(0.156)
0.036
(0.166)
0.300***
(0.144)
-0.087
(0.075)
0.034***
(0.016)
-0.000***
(0.000)
-0.846***
(0.190)
-0.244***
(0.088)
-0.508***
(0.109)
-0.737***
(0.153)
0.033
(0.132)
-0.008
(0.012)
-0.000
(0.000)
0.517
(0.399)
-0.062
(0.184)
-0.024
(0.210)
0.228
(0.170)
-0.175***
(0.081)
0.030***
(0.014)
-0.000***
(0.000)
-0.645***
(0.177)
-0.214***
(0.088)
-0.476***
(0.122)
-0.616***
(0.165)
-0.064
(0.069)
-0.006
(0.013)
-0.000
(0.000)
0.465
(0.330)
0.043
(0.176)
0.042
(0.211)
0.369***
(0.182)
-0.166***
(0.071)
0.032***
(0.010)
-0.000***
(0.000)
-0.841***
(0.163)
-0.203***
(0.091)
-0.466***
(0.145)
-0.683***
(0.145)
-0.068
(0.091)
-0.004
(0.017)
-0.000
(0.000)
0.304
(0.480)
-0.067
(0.210)
-0.135
(0.248)
0.050
(0.182)
-0.144***
(0.063)
0.028***
(0.010)
-0.000***
(0.000)
-0.768***
(0.130)
-0.213***
(0.090)
-0.461***
(0.129)
-0.471***
(0.162)
0.010
(0.068)
-0.007
(0.012)
-0.000
(0.000)
0.514***
(0.269)
0.014
(0.145)
0.040
(0.179)
0.330***
(0.168)
-0.087
(0.079)
0.038***
(0.012)
-0.000***
(0.000)
-0.620***
(0.174)
-0.093
(0.105)
-0.279***
(0.145)
-0.330***
(0.154)
0.097
(0.076)
0.009
(0.012)
-0.000
(0.000)
0.362***
(0.189)
0.100
(0.113)
0.171
(0.134)
0.374***
(0.131)
-0.070
(0.109)
0.018***
(0.010)
-0.000
(0.000)
-0.775***
(0.285)
-0.066
(0.156)
-0.260
(0.184)
-0.555***
(0.155)
0.076
(0.111)
-0.002
(0.027)
-0.000
(0.000)
-0.156
(0.566)
-0.252***
(0.146)
-0.546***
(0.164)
-0.596***
(0.173)
-0.111
(0.070)
-0.241
(0.186)
-0.181
(0.183)
-0.171***
(0.069)
-0.207***
(0.094)
-0.079
(0.097)
-0.346***
(0.094)
-0.060
(0.081)
-0.029
(0.113)
0.089***
(0.053)
0.205
(0.137)
0.255
(0.202)
0.057
(0.083)
0.138
(0.101)
0.268***
(0.116)
0.279***
(0.091)
0.043
(0.075)
0.077
(0.062)
-0.129***
(0.068)
-0.265
(0.180)
-0.210
(0.183)
-0.158***
(0.065)
-0.205***
(0.096)
-0.122
(0.097)
-0.354***
(0.095)
-0.062
(0.081)
-0.049
-0.045
(0.054)
0.107
(0.087)
0.148
(0.219)
0.201***
(0.098)
0.203
(0.133)
0.111
(0.123)
0.340***
(0.103)
0.071
(0.118)
-0.017
-0.113
(0.079)
-0.179
(0.148)
-0.072
(0.199)
-0.221***
(0.058)
-0.184***
(0.104)
-0.080
(0.118)
-0.355***
(0.079)
-0.036
(0.062)
-0.053
-0.001
(0.068)
0.236***
(0.129)
0.270
(0.198)
0.090
(0.086)
0.212***
(0.092)
0.218***
(0.118)
0.247***
(0.112)
0.002
(0.104)
0.041
-0.103
(0.093)
-0.183
(0.130)
0.005
(0.219)
-0.128***
(0.071)
-0.124
(0.103)
-0.135
(0.089)
-0.318***
(0.069)
-0.056
(0.062)
-0.044
0.037
(0.089)
0.017
(0.178)
0.288
(0.303)
0.179***
(0.103)
0.210***
(0.122)
0.108
(0.110)
0.202
(0.163)
0.104
(0.121)
-0.004
-0.149
(0.095)
-0.135
(0.138)
-0.013
(0.206)
-0.192***
(0.073)
-0.128
(0.095)
-0.106
(0.091)
-0.308***
(0.074)
0.012
(0.074)
-0.089
-0.031
(0.064)
0.216***
(0.116)
0.293***
(0.133)
-0.033
(0.087)
0.040
(0.079)
0.176***
(0.099)
0.116
(0.118)
0.066
(0.076)
-0.001
-0.113
(0.099)
-0.151
(0.160)
0.029
(0.223)
-0.063
(0.055)
-0.161***
(0.085)
-0.074
(0.082)
-0.272***
(0.086)
0.029
(0.111)
-0.127
-0.034
(0.075)
0.054
(0.165)
0.161
(0.178)
0.036
(0.089)
0.010
(0.085)
0.107
(0.097)
0.131
(0.126)
0.018
(0.106)
0.001
-0.128***
(0.062)
-0.266***
(0.132)
-0.255
(0.216)
0.006
(0.094)
-0.103
(0.107)
-0.327***
(0.119)
-0.318***
(0.086)
0.010
(0.070)
-0.110
-0.099
(0.140)
-0.203
(0.216)
0.154
(0.332)
0.401***
(0.138)
0.270***
(0.151)
-0.204
(0.143)
0.093
(0.191)
0.281
(0.172)
-0.108
(0.116)
(0.076)
(0.084)
(0.070)
(0.096)
(0.105)
(0.082)
(0.088)
(0.115)
(0.106)
(0.080)
(0.185)
Relational Goods
Married
Divorced
Widowed
Has one child
Has two or more children
Consider religion to be important
Has friends
Has employment
Has a supervisor
(Table continued)
29
Table 7. Factors Associated with Feeling Middle Class, by Definition (continued)
Dependent Variable:
Subjective social ranking
Based on absolute thresholds
US$2-US$10 PPP a day
US$2-US$13 PPP a day
(1)
(2)
(1)
(2)
Based on percentiles
p20 - p80
p30 - p90
(1)
(2)
(1)
(2)
Based on the median
0.5*p50 -1.5*p50
0.75*p50 -1.25*p50
(1)
(2)
(1)
(2)
Mixed measures
US$10 PPP a day - p95
(1)
(2)
Material conditions of life
Income
Household's monthly per capita income, US$ PPP, logs
-0.173***
(0.045)
0.220***
(0.056)
-0.182***
(0.044)
0.264***
(0.086)
-0.181***
(0.046)
0.263***
(0.074)
-0.218***
(0.041)
0.259***
(0.097)
-0.177***
(0.037)
0.150***
(0.072)
-0.139***
(0.040)
0.139***
(0.062)
-0.226***
(0.053)
0.104
(0.125)
-0.130
(0.119)
-0.375***
(0.103)
-0.141
(0.105)
-0.199***
(0.055)
0.179***
(0.049)
0.308***
(0.077)
0.217***
(0.082)
0.229***
(0.042)
-0.149
(0.116)
-0.388***
(0.103)
-0.162
(0.107)
-0.212***
(0.054)
0.135***
(0.053)
0.339***
(0.107)
0.133
(0.090)
0.257***
(0.040)
-0.115
(0.104)
-0.417***
(0.094)
-0.105
(0.104)
-0.168***
(0.045)
0.124***
(0.058)
0.227***
(0.078)
0.179***
(0.094)
0.224***
(0.045)
-0.131
(0.088)
-0.363***
(0.076)
-0.298***
(0.081)
-0.189***
(0.027)
0.127***
(0.053)
0.188
(0.131)
-0.050
(0.101)
0.293***
(0.086)
-0.134
(0.092)
-0.310***
(0.077)
-0.281***
(0.087)
-0.180***
(0.036)
0.121***
(0.061)
0.233***
(0.080)
0.045
(0.114)
0.147***
(0.038)
-0.132
(0.082)
-0.237***
(0.080)
-0.150***
(0.088)
-0.166***
(0.033)
0.061
(0.075)
0.175***
(0.088)
0.173***
(0.066)
0.137***
(0.030)
-0.176***
(0.049)
-0.383***
(0.079)
-0.299***
(0.104)
-0.193***
(0.040)
0.019
(0.072)
-0.050
(0.155)
-0.274
(0.191)
0.252***
(0.096)
-0.322***
(0.072)
-0.070
(0.095)
-0.275***
(0.062)
-0.245***
(0.100)
-0.247***
(0.092)
-0.215***
(0.103)
-0.077
(0.089)
0.206***
(0.107)
0.133***
(0.078)
0.131
(0.148)
0.219***
(0.076)
0.205***
(0.046)
-0.313***
(0.074)
-0.082
(0.096)
-0.283***
(0.060)
-0.255***
(0.105)
-0.251***
(0.086)
-0.237***
-0.001
(0.087)
0.194
(0.121)
0.158***
(0.091)
0.027
(0.174)
0.246***
(0.062)
0.229***
-0.260***
(0.067)
-0.088
(0.113)
-0.254***
(0.060)
-0.288***
(0.103)
-0.301***
(0.091)
-0.168***
-0.067
(0.110)
0.123***
(0.071)
0.125
(0.089)
0.127
(0.162)
0.143***
(0.080)
0.155***
-0.202***
(0.059)
-0.096
(0.091)
-0.234***
(0.058)
-0.282***
(0.115)
-0.176***
(0.077)
-0.186***
0.021
(0.103)
0.038
(0.105)
0.290***
(0.095)
0.056
(0.242)
0.163
(0.103)
0.029
-0.194***
(0.054)
0.009
(0.084)
-0.276***
(0.056)
-0.317***
(0.090)
-0.163***
(0.093)
-0.183***
-0.006
(0.082)
0.292***
(0.083)
0.008
(0.076)
0.023
(0.115)
0.138***
(0.080)
0.145***
-0.122***
(0.073)
-0.034
(0.105)
-0.259***
(0.090)
-0.247***
(0.090)
-0.084
(0.084)
-0.176***
0.042
(0.094)
0.328***
(0.092)
-0.022
(0.082)
-0.001
(0.129)
0.147***
(0.085)
0.118
0.063
(0.086)
-0.218
(0.136)
-0.107
(0.079)
-0.266***
(0.123)
-0.211***
(0.089)
-0.216***
0.132
(0.143)
0.057
(0.231)
0.281***
(0.138)
-0.223
(0.394)
0.075
(0.134)
-0.006
(0.106)
(0.085)
(0.095)
(0.066)
(0.076)
(0.106)
(0.075)
(0.079)
(0.076)
(0.090)
(0.042)
(0.109)
Financial circumstances
Access to financial services index
Does not have shortage of income to cover food costs
Does not have shortage of income to cover housing costs
Not concerned with financial matters
Physical Goods
Owns a house
Access to running water service
Access to telephone service
Has a television
Has a computer
Has an automobile
(Table continued)
30
Table 7. Factors Associated with Feeling Middle Class, by Definition (continued)
Dependent Variable:
Subjective social ranking
Based on absolute thresholds
US$2-US$13 PPP a day
US$2-US$10 PPP a day
(1)
(2)
(1)
(2)
Based on percentiles
p20 - p80
p30 - p90
(1)
(2)
(1)
(2)
Based on the median
0.5*p50 -1.5*p50
0.75*p50 -1.25*p50
(1)
(2)
(1)
(2)
Mixed measures
US$10 PPP a day - p95
(1)
(2)
Physical Goods
Has washing machine
Has a freezer
Lives in urban area
Constant
Observations
Pseudo R 2
Log Likelihood ln L(β)
-0.121
(0.082)
-0.222***
(0.088)
-0.284***
(0.148)
0.162***
(0.078)
0.117
(0.077)
0.005
(0.112)
-0.136
(0.086)
-0.238***
(0.084)
-0.286***
(0.149)
0.166
(0.112)
0.089
(0.077)
-0.031
(0.132)
-0.138***
(0.063)
-0.233***
(0.094)
-0.175
(0.126)
0.155
(0.125)
0.110***
(0.061)
0.082
(0.142)
-0.153***
(0.053)
-0.242***
(0.063)
-0.310***
(0.112)
0.085
(0.164)
0.061
(0.097)
0.045
(0.150)
-0.112
(0.070)
-0.191***
(0.067)
-0.284***
(0.113)
0.201***
(0.092)
0.088
(0.086)
0.037
(0.119)
-0.027
(0.064)
-0.099
(0.060)
-0.293***
(0.105)
0.151
(0.110)
0.197***
(0.106)
0.030
(0.125)
-0.168***
(0.065)
-0.147***
(0.086)
-0.042
(0.114)
0.050
(0.129)
0.079
(0.140)
0.016
(0.161)
1.401***
(0.491)
-3.774***
(0.398)
1.413***
(0.488)
-4.395***
(0.480)
1.976***
(0.493)
-4.299***
(0.616)
2.843***
(0.476)
-4.459***
(0.941)
2.587***
(0.385)
-2.798***
(0.449)
2.589***
(0.437)
-2.111***
(0.465)
4.607***
(0.386)
-2.200***
(1.231)
8,613
0.149
-7000.5
8,613
0.151
-6365.2
8,613
0.097
-7294.9
8,613
0.107
-6864.9
Source: Authors’ calculations based on Gallup (2007)
Significant at: ***99%, **95%, *90% . Country dummies are not reported.
Notes: (1) Subjective Poor vs. Subjective Middle Class; (2) Subjective Rich vs. Subjective Middle Class.
31
8,613
0.097
-8417.3
8,613
0.091
-8361.6
8,613
0.149
-5068.4
Argentina
Peru
Chile
32
Peru
f) $US 2 PPP to $US 13 PPP a day
$400
$300
$200
$200
$100
$400
$0
$200
Costa Rica
Uruguay
Argentina
Dominican Rep.
Chile
Brazil
Mexico
Paraguay
Panama
Nicaragua
Honduras
Guatemala
El Salvador
Bolivia
Peru
Ecuador
El Salvador
$0
Ecuador
c) 20th to 80th percentiles
$700
$600
$400
$300
$200
$100
$0
Argentina
Uruguay
Costa Rica
Chile
Brazil
Dominican Rep.
Panama
Paraguay
Honduras
Mexico
Nicaragua
Guatemala
Ecuador
Bolivia
Peru
El Salvador
$100
Bolivia
$200
Argentina
Costa Rica
Uruguay
Brazil
Chile
Panama
Dominican Rep.
Paraguay
Honduras
Nicaragua
Mexico
Guatemala
Ecuador
Bolivia
El Salvador
Peru
b) 0.75-1.25 times the median
Guatemala
$200
Honduras
$300
Dominican Rep.
Peru
Bolivia
El Salvador
Ecuador
Guatemala
Mexico
Honduras
Nicaragua
Paraguay
Dominican Rep.
Panama
Chile
Brazil
$400
Mexico
Nicaragua
Panama
Paraguay
$0
Brazil
$100
Uruguay
$300
Argentina
e) $US 2 PPP to $US 10 PPP a day
Costa Rica
Ecuador
Uruguay
$0
El Salvador
Argentina
$100
Costa Rica
$300
Bolivia
a) 0.5 – 1.5 times the median
Guatemala
$400
Dominican Rep.
Peru
El Salvador
Ecuador
Bolivia
Guatemala
Honduras
Mexico
Nicaragua
$500
Mexico
Honduras
Nicaragua
Paraguay
Panama
Chile
Brazil
Uruguay
Costa Rica
Dominican Rep.
Paraguay
Panama
Chile
Brazil
Costa Rica
Uruguay
Argentina
Maximum, Minimum and Median of Monthly Household Income by Country
Figure 1. Household Income of the Middle Class
d) 30th to 90th percentiles
$1,000
$500
$800
$600
$400
$200
$0
g) $US 10 PPP to 95th percentile
$1,200
$1,000
$800
$600
Note: The horizontal line represents the Latin-American median per capita income, 2007 US$ PPP, for the middle class, according to each definition.
Source: Authors’ calculations based on Gallup (2007).
Appendix Table A.1. Marginal Effects of the Subjective Social Ranking Equation
Ordered Logit Estimation
Dependent Variable:
Subjective social ranking (wp5722, 0-10
scale)
Marginal Effects by subjective wealth scale
The poorest
1
2
3
4
5
6
7
8
9
The richest
-0.002***
(0.001)
0.000***
(0.000)
-0.000***
(0.000)
-0.014***
(0.004)
-0.003***
(0.001)
-0.006***
(0.001)
-0.008***
(0.001)
-0.004***
(0.002)
0.001***
(0.000)
-0.000***
(0.000)
-0.030***
(0.007)
-0.007***
(0.003)
-0.012***
(0.004)
-0.016***
(0.003)
-0.007***
(0.004)
0.002***
(0.000)
-0.000***
(0.000)
-0.055***
(0.013)
-0.012***
(0.005)
-0.023***
(0.006)
-0.032***
(0.005)
-0.012***
(0.006)
0.003***
(0.000)
-0.000***
(0.000)
-0.088***
(0.023)
-0.020***
(0.008)
-0.037***
(0.011)
-0.057***
(0.010)
-0.004***
(0.002)
0.001***
(0.000)
-0.000***
(0.000)
-0.029***
(0.007)
-0.007***
(0.003)
-0.015***
(0.006)
-0.032***
(0.008)
0.015***
(0.008)
-0.004***
(0.001)
0.000***
(0.000)
0.113***
(0.029)
0.025***
(0.010)
0.046***
(0.014)
0.064***
(0.010)
0.008***
(0.004)
-0.002***
(0.000)
0.000***
(0.000)
0.058***
(0.014)
0.013***
(0.005)
0.025***
(0.008)
0.043***
(0.010)
0.004***
(0.002)
-0.001***
(0.000)
0.000***
(0.000)
0.029***
(0.007)
0.007***
(0.002)
0.013***
(0.004)
0.023***
(0.005)
0.002
(0.001)
-0.000***
(0.000)
0.000***
(0.000)
0.013***
(0.004)
0.003***
(0.001)
0.006***
(0.002)
0.011***
(0.003)
0.000***
(0.000)
-0.000***
(0.000)
0.000***
(0.000)
0.002***
(0.001)
0.000***
(0.000)
0.001***
(0.000)
0.002***
(0.001)
0.000
(0.000)
-0.000***
(0.000)
0.000***
(0.000)
0.002***
(0.001)
0.000***
(0.000)
0.001***
(0.000)
0.002***
(0.001)
-0.002***
(0.001)
-0.003***
(0.001)
-0.003***
(0.002)
-0.002***
(0.001)
-0.003***
(0.001)
-0.002***
(0.001)
-0.006***
(0.001)
-0.001
(0.001)
-0.000
(0.001)
-0.004***
(0.001)
-0.006***
(0.002)
-0.007
(0.004)
-0.005***
(0.002)
-0.006***
(0.003)
-0.005***
(0.003)
-0.013***
(0.003)
-0.002
(0.002)
-0.000
(0.002)
-0.008***
(0.002)
-0.011***
(0.005)
-0.013***
(0.007)
-0.008***
(0.003)
-0.010***
(0.005)
-0.009***
(0.005)
-0.024***
(0.005)
-0.003
(0.003)
-0.000
(0.004)
-0.013***
(0.004)
-0.018***
(0.008)
-0.022***
(0.013)
-0.014***
(0.005)
-0.017***
(0.009)
-0.014***
(0.008)
-0.034***
(0.008)
-0.005
(0.005)
-0.000
(0.006)
-0.004***
(0.001)
-0.008***
(0.004)
-0.010
(0.007)
-0.005***
(0.002)
-0.006***
(0.004)
-0.004***
(0.002)
-0.006***
(0.001)
-0.002
(0.002)
-0.000
(0.002)
0.016***
(0.005)
0.023***
(0.009)
0.027***
(0.015)
0.017***
(0.006)
0.021***
(0.011)
0.018***
(0.010)
0.046***
(0.011)
0.007
(0.007)
0.001
(0.008)
0.008***
(0.002)
0.013***
(0.006)
0.015
(0.010)
0.009***
(0.003)
0.011***
(0.006)
0.009***
(0.005)
0.021***
(0.004)
0.003
(0.003)
0.000
(0.004)
0.004***
(0.001)
0.006***
(0.003)
0.008
(0.005)
0.005***
(0.002)
0.006***
(0.003)
0.004***
(0.003)
0.010***
(0.002)
0.002
(0.002)
0.000
(0.002)
0.002***
(0.001)
0.003***
(0.001)
0.004
(0.003)
0.002***
(0.001)
0.003***
(0.001)
0.002
(0.001)
0.005***
(0.001)
0.001
(0.001)
0.000
(0.001)
0.000***
(0.000)
0.000***
(0.000)
0.001
(0.000)
0.000***
(0.000)
0.000***
(0.000)
0.000***
(0.000)
0.001***
(0.000)
0.000
(0.000)
0.000
(0.000)
0.000***
(0.000)
0.000***
(0.000)
0.001
(0.000)
0.000***
(0.000)
0.000***
(0.000)
0.000
(0.000)
0.001***
(0.000)
0.000
(0.000)
0.000
(0.000)
Capabilities
Female
Age (years)
Age squared
Health score (EQ - 5D)
Complete primary education
Complete secondary education
Complete superior education
Relational Goods
Married
Divorced
Widowed
Has one child
Has two or more children
Consider religion to be important
Has friends
Has employment
Has a supervisor
33
(Table continued)
Appendix Table A.1. Marginal Effects of the Subjective Social Ranking Equation (continued)
Dependent Variable:
Subjective social ranking (wp5722, 0-10
scale)
Marginal Effects by subjective wealth scale
The poorest
1
2
3
4
5
6
7
8
9
The richest
-0.004***
(0.001)
-0.009***
(0.002)
-0.017***
(0.003)
-0.027***
(0.004)
-0.009***
(0.002)
0.034***
(0.006)
0.017***
(0.003)
0.009***
(0.002)
0.004***
(0.001)
0.001***
(0.000)
0.001***
(0.000)
-0.003***
(0.001)
-0.005***
(0.002)
-0.010***
(0.003)
-0.016***
(0.005)
-0.005***
(0.002)
0.021***
(0.007)
0.011***
(0.004)
0.005***
(0.001)
0.002***
(0.001)
0.000***
(0.000)
0.000***
(0.000)
-0.007***
(0.001)
-0.014***
(0.002)
-0.025***
(0.004)
-0.038***
(0.006)
-0.010***
(0.002)
0.050***
(0.008)
0.025***
(0.004)
0.012***
(0.002)
0.005***
(0.001)
0.001***
(0.000)
0.001***
(0.000)
-0.004***
(0.002)
-0.004***
(0.000)
-0.008***
(0.004)
-0.009***
(0.001)
-0.015***
(0.006)
-0.016***
(0.002)
-0.023***
(0.009)
-0.026***
(0.003)
-0.006***
(0.002)
-0.009***
(0.002)
0.031***
(0.012)
0.033***
(0.005)
0.015***
(0.006)
0.017***
(0.002)
0.007***
(0.003)
0.009***
(0.001)
0.003***
(0.001)
0.004***
(0.001)
0.000***
(0.000)
0.001***
(0.000)
0.000***
(0.000)
0.001***
(0.000)
-0.002***
(0.001)
-0.005***
(0.002)
-0.004***
(0.001)
-0.004***
(0.001)
-0.004***
(0.001)
-0.003***
(0.001)
-0.004***
(0.002)
-0.009***
(0.004)
-0.007***
(0.001)
-0.009***
(0.003)
-0.007***
(0.001)
-0.006***
(0.002)
-0.008***
(0.004)
-0.017***
(0.006)
-0.014***
(0.003)
-0.016***
(0.006)
-0.014***
(0.003)
-0.011***
(0.004)
-0.013***
(0.007)
-0.025***
(0.009)
-0.022***
(0.004)
-0.024***
(0.008)
-0.023***
(0.005)
-0.018***
(0.007)
-0.004***
(0.002)
-0.005***
(0.001)
-0.007***
(0.002)
-0.005***
(0.001)
-0.009***
(0.003)
-0.007***
(0.003)
0.017***
(0.009)
0.033***
(0.012)
0.028***
(0.006)
0.033***
(0.011)
0.029***
(0.006)
0.022***
(0.009)
0.008***
(0.004)
0.015***
(0.005)
0.014***
(0.003)
0.015***
(0.005)
0.016***
(0.004)
0.012***
(0.005)
0.004***
(0.002)
0.008***
(0.003)
0.007***
(0.002)
0.007***
(0.002)
0.008***
(0.002)
0.006***
(0.002)
0.002***
(0.001)
0.003***
(0.001)
0.003***
(0.001)
0.003***
(0.001)
0.004***
(0.001)
0.003***
(0.001)
0.000***
(0.000)
0.001***
(0.000)
0.001***
(0.000)
0.001***
(0.000)
0.001***
(0.000)
0.000***
(0.000)
0.000***
(0.000)
0.001***
(0.000)
0.000***
(0.000)
0.001***
(0.000)
0.001***
(0.000)
0.000***
(0.000)
Material conditions of life
Income
Household's monthly per capita income, US$
PPP, logs
Financial circumstances
Access to financial services index
Does not have shortage of income to cover food
costs
Does not have shortage of income to cover
housing costs
Not concerned with financial matters
Physical Assets
Owns a house
Access to running water service
Access to telephone service
Has a television
Has a computer
Has an automobile
(Table continued)
34
Appendix Table A.1. Marginal Effects of the Subjective Social Ranking Equation (continued)
Dependent Variable:
Subjective social ranking (wp5722, 0-10
scale)
Marginal Effects by subjective wealth scale
The poorest
1
2
3
4
5
6
7
8
9
The richest
-0.002***
(0.001)
-0.004***
(0.001)
-0.004***
(0.002)
-0.005***
(0.002)
-0.008***
(0.002)
-0.008***
(0.004)
-0.009***
(0.004)
-0.015***
(0.005)
-0.014***
(0.008)
-0.015***
(0.006)
-0.023***
(0.007)
-0.022***
(0.012)
-0.005***
(0.002)
-0.007***
(0.002)
-0.006***
(0.004)
0.019***
(0.008)
0.030***
(0.010)
0.028***
(0.016)
0.010***
(0.004)
0.015***
(0.004)
0.014***
(0.008)
0.005***
(0.002)
0.007***
(0.002)
0.007***
(0.004)
0.002***
(0.001)
0.003***
(0.001)
0.003***
(0.002)
0.000***
(0.000)
0.001***
(0.000)
0.000***
(0.000)
0.000
(0.000)
0.001***
(0.000)
0.000***
(0.000)
Physical Assets
Has washing machine
Has a freezer
Lives in urban area
Significant at: ***99%, **95%, *90%. Country dummies are not reported.
Source: Authors’ calculations based on Gallup (2007).
35